Run one synthetic leasing journey.
In a sandbox, have a made-up prospect ask about a unit after hours, book a tour, reschedule it, then ask for a human to confirm a detail the system can’t answer. Run the same scenario once through a listing marketplace and once through a direct property inquiry. Don’t use real applicant data or send real messages.
- Follow the record. Does the same prospect and appointment stay linked across the tools? Is the old slot freed?
- Follow the responsibility. Does one clearly assigned person get the request and the context to act, without duplicate follow-ups? Time the first substantive human response, not the automated acknowledgment.
- Check the renter’s view. Does the confirmation reflect the actual appointment? Does the renter know what happens next?
Pass: one prospect, one current appointment, and one named owner for the exception. If one of those is missing, solve that before scaling.